Cystic fibrosis
The story of the genetic disease cystic fibrosis (CF) is related to many of the technologies described in this issue of the IBM Systems Journal.^1-5 CF is one of the most common fatal genetic diseases in developed countries, affecting about 30,000 people in the U.S. and 25,000 in Europe. Affected individuals have abnormally thick lung secretions and frequently succumb to respiratory infections.

This disease is caused by a defect in a gene on chromosome 7 that codes for a protein known as CFTR, which consists of 1480 amino acids. The CFTR gene was identified in 1989 by a large group of collaborators. The CFTR protein plays a critical role in epithelial cells, which often function in the secretion of mucous and other fluids. CFTR is a transmembrane protein, embedded in the cell membrane, where it mediates the transport of chloride ions and water molecules. The most common defect to the CFTR gene is the omission of a single phenylalanine amino acid at position 508 in the CFTR protein. It is currently thought that the missing amino acid results in a misfolding of the CFTR protein, at which point it is degraded by normal cellular housekeeping. As a result, no CFTR protein finds its way to the cell membrane, disrupting normal cell function.

A cure for CF rests partially on the full characterization of the structure, folding mechanism, and function of the CFTR protein. However, transmembrane proteins such as CFTR are nearly impossible to crystallize, preventing X-ray crystallography experiments that would reveal its 3-D structure. Structure-prediction methods based on similarity of a protein's sequence to known structures are only partially revealing. A structure for a 214-amino-acid subregion of CFTR, known as NBD1 (nucleotide binding domain one), based on purely computational approaches, was reported in 1996. This domain includes amino acid 508, which is missing in most CF patients. The rightmost ribbon diagram above is a skeletal depiction of the subregion's structure.

Nuclear magnetic resonance (NMR) techniques can be used to determine detailed structural information for short peptide sequences. Examples of NMR-determined CFTR peptide structures are also shown; they are two short peptides of 25 and 26 amino acids. The larger one includes, and the smaller one omits, the phenylalanine that is missing from position 508 in CF patients. The red regions in the peptide and in the theoretical NBD1 structure represent this amino acid. Corresponding regions before and after position 508 are shown in yellow (498 to 507) and green (509 to 523). These NMR structures are somewhat at odds with the theoretical structure shown for NBD1. However, the surrounding material in the larger domain may serve to stabilize different structures than are exhibited by short peptides in solution. Closing the gap between theoretical and actual structure will be the first step to understanding and conquering CF.

In February 2001, the Cystic Fibrosis Foundation announced a collaboration with Structural GenomiX involving "11 million to fund a five-year project to produce a full 3-D structure for the CFTR protein. It is hoped that this structural information will be useful to developing an understanding of how the protein works and in the design of new treatments.

AIDS
The story of the infectious disease AIDS is also related to many of the technologies described in this issue.^6-9 AIDS (acquired immunodeficiency syndrome) is one of the most devastating diseases in modern human existence. It is caused by the human immunodeficiency virus (HIV). HIV weakens the immune system, making the body unable to resist ordinarily benign infections. HIV, like most viruses, takes advantage of the cellular machinery of the cells it infects to make the proteins needed for its replication. These proteins are coded in the genetic material of the invading virus. Several HIV proteins are synthesized in one continuous chain, known as a polyprotein. The polyproteins are then cleaved into smaller chains that are reassembled into new viruses. The cleavage of the polyprotein is performed by a viral enzyme known as a protease.

In 1988, the amino acid sequence of HIV-1 protease and its role in the replication of the virus was established. The X-ray crystallographic structure of HIV-1 protease was determined in 1989, yielding its 3-D shape. The HIV-1 protease structure is shown above. HIV-1 protease is itself a symmetric protein, a dimer composed of two identical strands.

The HIV-1 protease works by binding to the HIV polyprotein and cleaving it at specific places. Using the tools of rational drug design, a number of drugs have been designed that bind strongly to the active site in HIV-1 protease, thereby interfering with its ability to cleave the polyprotein. Such drugs are known as HIV-1 protease inhibitors. Protease inhibitors have proved to be one of the most effective classes of drugs for the treatment of HIV infections. A second highly effective class of drugs is known collectively as reverse transcriptase inhibitors (AZT is an important example). Drugs from each of the two classes are often delivered together in the form of combination, or cocktail, therapy.

In rational drug design, chemists make extensive use of computers along with graphical representations of the 3-D molecular structure of the protease, as determined by X-ray crystallographic experiments, to design molecules that bind more strongly to the active site than the polyprotein. Not only is the shape of the protease used in this process, but also computational predictions of properties such as solubility and toxicity of the candidate molecules. Dozens of HIV-1 protease inhibitors have been designed using these techniques, including saquinavir (Invirase, from Hoffman-LaRoche, shown here bound to HIV-1 protease), indinavir (Crixivan, from Merck), nelfinavir (Viracept, from Agouron), and ritonavir (Norvir, from Abbott). With these kinds of experimental and computational advances, we may some day soon have HIV under control as a treatable disease.

William C. Swope

Cited references and notes

1. For information about the Protein Data Bank, from which the molecular diagrams shown on the cover and on these two pages were developed, see: H. M. Berman et al., "The Protein Data Bank," Nucleic Acids Research 28, 235-242 (2000).
2. PBD IB: 1NBD. F. J. Hoedemaeker et al., "A Model for the Nucleotide-Binding Domains of ABC Transporters Based on the Large Domain of Aspartate Aminotransferase," Proteins: Structure, Function, and Genetics 30, 275-286 (1998).
3. PDB IDs: 1CKW, 1CKX, 1CKY, and 1CKZ. M. A. Massiah et al., "Cystic Fibrosis Transmembrane Conductance Regulator: Solution Structures of Peptides Based on the Phe508 Region, the Most Common Site of Disease-Causing F508 Mutation," Biochemistry 38, No. 23, 7453-7461 (1999).
4. D. M. Orenstein, Cystic Fibrosis, A Guide for Patient and Family, Second Edition, Lippincott-Raven Publishers, Philadelphia, PA (1997).
5. M. J. Welsh and A. E. Smith, "Cystic Fibrosis," Scientific American 273, 52-59 (1995).
6. PDB ID: 1HXB, A. Krohn et al., "Novel Binding Mode of Highly Potent HIV-Proteinase Inhibitors Incorporating the (R)-Hydroxyethylamine Isostere," Journal of Medicinal Chemistry 34, No. 11, 3340-3342 (1991).
7. P. Y. Lam et al., "Rational Design of Potent, Bioavailable, Nonpeptide Cyclic Ureas as HIV Protease Inhibitors," Science 263, pp. 380-384 (1994).
8. The National Cancer Institute maintains an HIV protease database available over the Internet at the Web site http://www.ncifcrf.gov/CRYS/HIVdb.
9. Some of the material on HIV-1 protease was adopted from the highly informative Web site on rational drug design by D. Eric Walters of the Chicago Medical School: http://www.finchcms.edu/biochem/walters/walters_lect/walters_lect.html.